Misinterpreted myoepithelial carcinoma of salivary gland: a challenging and potentially significant pitfall

Abstract

Myoepithelial carcinoma (MECA) is an underrecognized challenging entity with a broad morphologic spectrum. Misinterpreting MECA is not uncommon as distinguishing it from its mimics, especially cellular myoepithelial-rich pleomorphic adenoma (PA), can be difficult. We described 21 histologically challenging cases of MECAs (16 MECA ex-PA and 5 MECA de novo). All MECAs ex-PA were intracapsular or minimally invasive except for three cases. Eighteen (86%) were initially misinterpreted as benign neoplasms, including PA (10), atypical PA (5), and myoepithelioma (3). The remaining three were initially diagnosed as malignant (MECA ex-PA) but were histologically challenging. Histologic features that were found most helpful in recognizing the malignant nature of MECA included: uniformly cellular myoepithelial proliferation with an expansile nodular lobulated pattern (all cases) and alternate hypocellular and hypercellular zonal distribution (76% of cases). Among the 16 MECA patients with follow up, 14 (87.5%) progressed: 10 developed local recurrence and 5 distant metastases. In contrast, only one of 33 patients with cellular PA (control group) recurred locally. Ten of the 14 MECAs that progressed were MECA ex-PA, and 12 (85%) had an initial benign diagnosis. Two patients with MECA ex-PA died of their disease; one had an initial diagnosis of PA. MECA is a histologically challenging entity that closely mimics PA and seems to carry a significant risk of recurrence. Areas of clonal appearing cellular myoepithelial growth with an expansile nodular lobulated pattern and zonal cellular distribution distinguish the majority of MECAs and may serve as useful diagnostic histologic features to differentiate MECA from its benign mimics.

Introduction

Salivary gland neoplasms are generally rare and histologically diverse, with 20 distinct types being recognized by the World Health Organization (WHO) classification of head and neck tumors (4^th edition, 2017) (1). Among salivary gland tumors, myoepithelial carcinoma (MECA) seems particularly very challenging, given the fact that it is rare and understudied with poorly defined diagnostic criteria. In earlier studies, MECA was reported to account for less than 2% of all salivary gland malignancies, but the incidence of MECA is currently believed to be higher. This is mainly due to the fact that MECA is easily under-recognized given its broad histologic spectrum and its morphologic overlap with other benign or malignant salivary gland tumors (2, 3). MECA may arise in association with pre-existing PA (MECA ex-PA) or de novo, and may affect major or minor salivary glands (1). Histologically, MECA is defined as a tumor that is composed almost exclusively of myoepithelial cells and characterized by invasive growth pattern. When morphologic features of malignancy such as infiltrative destructive growth or overt cytologic atypia with high grade features are obvious, the malignant nature of the tumor can be readily recognized. Nonetheless, MECA often appears, histologically and cytologically, exceptionally bland. In such cases, the diagnosis may be easily overlooked and misclassified as a benign salivary gland neoplasm, such as cellular or myoepithelial-rich PA. This is in particular challenging in cases of MECA ex-PA, especially if the tumor shows minimal or no capsular invasion (3, 4). Clinically, MECA seems to be relatively aggressive even when it is intracapsular or minimally invasive MECA ex-PA (4, 5); it is associated with a high frequency of recurrence, including 37% risk of local recurrence and 22% chance of distant metastasis (3, 4, 6–9). Therefore, it is important to distinguish MECA from its benign mimics at the early stage of the disease, to provide adequate management and follow up for the patients.

In this study, we reported 21 cases of diagnostically challenging MECAs in which a benign diagnosis was initially rendered or strongly considered. We also compared these cases to a control group of 33 cellular PAs with the attempt to determine the histopathologic features that can help establish the correct diagnosis of MECA and distinguish it from benign neoplasms.

Material and methods

Inclusion criteria and study cohorts:

After obtaining Institutional Review Board approval, the pathology database was searched for cases of cellular PA, atypical PA and MECA diagnosed at Memorial Sloan-Kettering Cancer Center (MSKCC, New York, NY, USA) and Sunnybrook Health Sciences Centre (SHSC, Toronto, ON, Canada) between 1999 and 2017. A total of 21 histologically challenging cases of MECAs (MSKCC: n=20 and SHSC: n=1), either de novo or ex-PA, were included in the present study. Three of these cases were included in our previous studies (3, 10, 11). All cases were reviewed by at least four pathologists (BX, RG, WM and NK) and a consensus diagnosis of MECA was reached in all cases. In addition, a control group of 33 cellular PAs (MSKCC: n=26 and SHSC: n=7) was also included in order to compare their histologic features and clinical outcomes with MECA cases. Cellular PA was defined as tumors in which the epithelial or myoepithelial component accounted for more than 50% of total tumor volume.

Histopathologic review and ancillary studies:

The following histologic features of the MECAs were reviewed: presence or absence of PA component (MECA ex-PA versus MECA de novo), nuclear pleomorphism, mitotic index, tumor necrosis, margin status, lymphovascular invasion, perineural invasion, tumor architecture including nodular lobulated expansile growth pattern and presence of alternate hypocellular and hypercellular areas (i.e. zonal cellular distribution), hyalinized stroma, and squamous metaplasia. In addition, for cases of MECA ex-PA, the percentage of myoepithelial carcinoma, the presence or absence of capsular invasion, as well as the extent of capsular invasion using the proposed criteria in the WHO classification (4^th edition) were also assessed. Site of primary tumor, tumor size and lymph node status (in cases with lymph node sampling) were recorded. Tumor sampling for histopathology was reported in 18 cases. The initial diagnoses of the 21 MECA cases were documented. Mitotic index was determined by counting 10 high-power fields (HPFs, 400X, total field size 2.4 mm²) with an Olympus microscope (U-DO model BX41, Olympus America Inc., Center Valley, PA, United States) in the areas of highest concentration of mitotic figures. Ki67 proliferation index was performed using a MIB-1 rabbit monoclonal antibody (clone 30-9, dilution 1: 200, Ventana Medical Systems Inc. Tucson, AZ, US) and documented in 8 cases, including one MECA de novo and 7 MECAs ex-PA. Twelve MECAs were tested by immunohistochemistry including cytokeratin AE1/AE3 (n=7, DAKO M3515 clone, dilution 1.1600), CAM5.2 (n=5, BD Bioscience 349205, dilution 1.75), S100 (n=9, DAKO Z0311 clone, dilution 1:8000), calponin (n=11, Ventana Medical System Inc pre-diluted EP789Y clone), smooth muscle actin (n=11, Vector laboratory asm-1, dilution 1:50) and p63 (n=9, Ventana pre-diluted mouse monoclonal antibody 4A4). Molecular analysis was performed in three cases, including florescence in situ hybridization (FISH) for PLAG1 and massive parallel high output sequencing using MSKCC-IMPACT™ (Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets). Detailed descriptions of both techniques (MSKCC-IMPACT and PLAG1 FISH) have been previously reported by our group (10, 12, 13).

Clinical review:

The patients’ charts were reviewed for age at diagnosis, gender, site of primary tumor, recurrence, and outcomes. Follow up data were available on 16 patients with MECA and 19 patients with cellular PA. Recurrence was determined on the basis of clinical, radiologic and pathologic examination, and was further classified as local recurrence and distant metastasis.

Statistics:

All statistical analyses were performed using the SPSS software 24.0 (IBM Corporation, New York, NY, USA). The clinical features were compared between MECAs and cellular PAs using appropriate statistical tests, i.e. Fisher’s exact test for nonparametric variables and two-tailed Student’s t test for continuous variables. Disease-specific survival and disease-free survival were calculated from the date of surgery to the date of disease-related death or first recurrence and were compared between MECAs and PAs using the Log rank test. P values less than 0.05 were considered to be statistically significant.

Results

The initial diagnoses of MECAs

Eighteen out of 21 (86%) MECAs had an initial diagnosis of benign neoplasms, including 10 PAs, 5 atypical PAs, and 3 myoepitheliomas. The other three cases were initially diagnosed as malignant (MECA ex-PA) but were included in the study because they were believed to be histologically challenging and the diagnosis of PA was strongly considered.

Histopathologic features of MECAs

The primary site of MECAs included parotid (n=14, 67%), submandibular (n=5, 24%) and minor salivary glands (n=2, 9%). The histologic features of MECAs are summarized in Table 1. Among the 21 MECAs, 16 were MECAs ex PA and 5 were MECAs de novo. The majority of MECAs ex-PA (with the exception of three cases) were intracapsular (10/16, 62%) or minimally invasive (3/16, 19%). The PA component showed marked hyalinized stroma in 5 out 14 cases. The tumors had a mitotic index that ranges from 0-9 mitoses per 10 high power fields (HPFs) (median=2/10 HPFs). Five out of 21 tumors (24%) exhibited tumor necrosis. Nuclear pleomorphism was noted in 5 out of 21 tumors (24%), four of which had only focal pleomorphism. Squamous metaplasia was identified in 3 cases (14%). All 21 MECAs showed a predominant uniform cellular proliferation of myoepithelial cells with an invasive expansile nodular lobulated pattern within the tumor. Of the 21 cases, 16 (76%) exhibited alternate hypocellular and hypercellular areas of myoepithelial cells (i.e. zonal cellular distribution) within the expansile nodules (Figures 1 and 2). In MECAs ex-PA, the nodular cellular myoepithelial growth was seen intermixed with the PA component in 11 out of 16 cases (69%). No lymphovascular or perineural invasion was seen. Among the 17 cases with available margin status, 6 (35%) had positive margin and 11 (65%) had negative margin.

Table 1.

Clinicopathologic features of the 21 studied MECAs.

		MECA (n = 21)
Sex	Female	12 (57%)
	Male	9 (43%)
Age, mean, median (range)		59, 66 (28-88)
Tumor size cm, mean, median (range)		3.4, 3.0 (0.9-7.9)
Primary site	Parotid	14 (67%)
	Submandibular	5 (24%)
	Minor salivary glands	2 (9%)
Nodal status	N0	6 (29%)
	Nx	15 (71%)
Initial diagnosis	PA	10 (48%)
	Myoepithelioma	3 (14%)
	Atypical PA	5 (24%)
	MECA ex-PA	3 (14%)
Nodular lobulated growth	Present	21 (100%)
Zonal distribution	Absent	5 (24%)
	Present	16 (76%)
MI (/10HPFs), mean, median (range)		2, 2 (0-9)
MECA de novo vs. MECA ex-PA	MECA ex-PA	16 (76%)
	MECA de novo	5 (24%)
Extent of invasion in MECA ex-PA (n=16)	Intracapsular	10 (63%)
	Minimally invasive	3 (19%)
	Invasive	3 (19%)
Necrosis	Absent	16 (76%)
	Present	5 (24%)
Squamous metaplasia	Absent	18 (86%)
	Present	3 (14%)
Hyalinized stroma	Absent	15 (71%)
	Focally present	1 (5%)
	Present	5 (24%)
Nuclear pleomorphism	Absent	16 (76%)
	Present	5 (24%)
LVI	Absent	21 (100%)
PNI	Absent	21 (100%)
Margin status (n=17)	Negative	11 (65%)
	Positive	6 (35%)
FU period month, mean, median (range)		54, 47 (7-132)
Status at last follow up (n = 16)	NED	7 (44%)
	AWD	7 (44%)
	DOD	2 (12%)
Recurrence	No	2 (12.5%)
	Yes	14 (87.5%)
Local recurrence	No	6 (38%)
	Yes	10 (63%)
Distant metastasis	No	11 (69%)
	Yes	5 (31%)

MECA: myoepithelial carcinoma; MECA ex-PA: myoepithelial carcinoma ex-pleomorphic adenoma; PA: pleomorphic adenoma; LVI: lymphovascular invasion; PNI: perineural invasion; MI: mitotic index; HPFs: high power fields; FU: follow up period; NED: no evidence of disease; AWD: alive with disease; DOD: dead of disease.

Figure 1. An intracapsular MECA ex-PA.

located in the parapharyngeal space in a 41-year-old female (A to C). The tumor was initially diagnosed as atypical PA. Histologically, the tumor exhibited expansile lobulated nodular growth with hypocellular (H) and hypercellular (HY) zonal arrangement (A). Typical area of PA with myxoid stroma and ductal and myoepithelial elements was present (B). The carcinoma component showed cellular myoepithelial growt3h with scattered mitoses (arrow). The tumor was positive for SMA, AE1/AE3, S100, and GFAP. (D) The patient developed a lytic sacral metastasis 4 year after the initial diagnosis (blue arrows). The metastatic MECA showed similar histology to the primary tumor (E and F).

MECA; myoepithelial carcinoma; PA: pleomorphic adenoma.

Figure 2. Histologic features of MECA ex-PA in three cases (A-C).

In all three cases, MECA component showed nodular lobulated myoepithelial growth, separate from the PA component. Zonal cellular distribution with central hypocellular and peripheral hypercellular growth was noted in tumors B and C. Inserts (400X): Epithelioid, spindle, or clear cell cytomorphology of MECA. Nuclear pleomorphism was noted in tumor A.

MECA: myoepithelial carcinoma; PA: pleomorphic adenoma.

Ancillary studies: immunohistochemistry profile and molecular alterations of MECAs

Immunohistochemical stains were available in 12 cases. The stained MECAs were positive for AE1/AE3 (7/7, 100%), CAM5.2 (4/5, 80%), S100 (9/9, 100%, diffuse staining), calponin (11/11, 100%), smooth muscle actin (9/11, 82%), and p63 (8/9, 89%). Eight cases had a Ki67 immunohistochemical staining available, ranging from <5% to 10%.

Molecular studies were performed in three cases, including PLAG1 FISH and next generation targeted sequencing using MSKCC-IMPACT platform™. PLAG1 fusion was identified on FISH analysis in a case of de novo MECA that did not show an obvious histologic evidence of a PA component in the entirely submitted tumor. CTNNB1-PLAG1 in-frame fusion between CTNNB1 exon 1 and PLAG1 exon 2 was detected on the IMPACT in a MECA ex-PA. EML4-ALK in frame fusion including the kinase domain of ALK, and p53 mutation were detected on IMPACT in a locally recurrent MECA ex-PA. In the latter case, the recurrent tumor was histologically high grade showing both ductal and myoepithelial differentiations.

Clinical outcome of MECAs

The clinical outcomes of patients with MECA are summarized in Table 2. Among the 16 MECA patients with available follow up (median follow up period of 47 months), 14 (88%) progressed, including 9 with local recurrence, 4 with distant metastases, and one with both local and distant recurrences. Twelve out of 14 (85%) patients with progression had an initial benign diagnosis, including 7 PAs, 3 myoepitheliomas, and 2 atypical PAs. Ten out the14 patients who progressed had MECA ex-PA (6 intracapsular, 3 minimally invasive, and one invasive) and 4 were de novo MECAs. Follow up was not available on one patient with MECA with necrosis. All the other three patients with MECA showing necrosis had recurrences (2 distant and one local).

Table 2.

Clinical comparison between MECA and cellular PA.

		MECA (n=21)	Cellular PA (n=33)	P value
Sex	Female	12 (57%)	17 (52%)	0.783
	Male	9 (43%)	16 (48%)
Age, mean, median, range		59, 66 (28-88)	49, 50 (14-80)	0.036
Tumor size cm, mean, median (range)		3.4, 3.0 (0.9-7.9)	2.4, 2.1 (1.1 - 7.5)	0.039
Site of primary tumor	Parotid	14 (67%)	29 (88%)	0.136
	Submandibular	5 (24%)	2 (6%)
	Minor salivary glands	2 (9%)	2 (6%)
Margin status	Negative	11 (52%)	28 (85%)	0.019
	Positive	6 (29%)	2 (6%)
	N/A	4 (19%)	3 (9%)
FU period month, mean, median (range)		54, 47 (7-132)	37, 23 (1-184)	0.205
Status at last follow up (MECA: n = 16; PA: n = 19)	NED	7 (44%)	19 (100%)	0.288
	AWD	7 (44%)	0 (0%)
	DOD	2 (12%)	0 (0%)
Recurrence	No	2 (12.5%)	18 (94%)	<0.001
	Yes	14 (87.5%)	1 (6%)
Distant metastasis	No	11 (69%)	19 (100%)	0.018
	Yes	5 (31%)	0 (0%)

Values are expressed as number of cases (percentage of column total) unless specified otherwise.

P values were obtained using Fisher’s exact tests for categorical variables, Student’s T test for continuous variables, and Log rank test for outcome and survival.

MECA; myoepithelial carcinoma; PA: pleomorphic adenoma; NED: no evidence of disease; AWD: alive with disease; DOD: dead of disease.

Among the 14 MECAs that progressed, 10 (7 local and 3 distant) demonstrated similar morphology in recurrent and primary tumors without evidence of histologic progression (8 had bland cytology and 2 had necrosis in both primary and recurrent tumors), while the remaining 4 cases (2 local, one distant and one local and distant) exhibited histologic progression showing high grade features, including necrosis, (n=3), increased mitotic index 5-10/HPFs (n=4), and nuclear pleomorphism (n=4) in the recurrent but not in the primary tumor. One locally recurrent high-grade tumor showed both ductal and myoepithelial differentiation.

Five patients developed distant metastasis 26 to 81 months after the initial resection of the primary salivary gland tumors. The site of distant metastases included lung (n=2), bone (n =2) and skin/soft tissue (n=1). The original diagnoses of the primary tumor in these five cases were PA (n=1), atypical PA (n=2), and intracapsular MECA ex-PA (n=2). Upon histologic review of these five cases at MSKCC, the three tumors diagnosed as PA and atypical PAs were re-classified as MECA ex-PA (2 intracapsular and one minimally invasive).

Two patients with MECA ex-PA died of their diseases. The first patient had a 4 cm intracapsular MECA ex-PA, developed lung metastasis 26 months after presentation and died of his disease, 29 months after his initial diagnosis. The second patient had a 5cm tumor (initially diagnosed as PA and reclassified as minimally invasive MECA ex-PA), suffered from a local recurrence, 16 months after diagnosis and died of uncontrollable locoregional failure, 45 months after the initial surgery.

The clinicopathologic comparison with cellular PAs

MECA seemed to affect older patients and displayed a larger tumor size as compared to cellular PA, with a median age at diagnosis of 66 and 50, respectively (p=0.036), and a median tumor size of 3.3 cm (range: 0.9-7.9 cm, p=0.039) and 2.4 cm (range: 1.1 to 7.5 cm) (p value), respectively. There was a trend of submandibular and minor salivary glands involvement and positive margins seen more frequently in MECAs than in cellular PAs, however this was not statistically significant (p=0.136 and 0.067, respectively). Most of the cellular PAs (88%, 29/33) were located in the parotid gland, while only a minority (12%) was located in the submandibular and minor salivary glands (6%, 2/33 in each location). In contrast, over a third of MECAs was identified in the submandibular and minor salivary glands (9%, 2/21 and 26%, 5/21 respectively). The margin status was not available in 3 cellular PAs and 4 MECAs. The margin was positive only in 2 out of 30 (7%) cellular PAs whereas six out of 17 (35%) MECAs had positive margin (p=0.019). No significant difference was found between the two groups in term of patients’ gender (p=0.783) and follow up period (p=0.288).

Histologically, all cellular PAs lacked the homogenous monotonous growth of myoepithelial cells, the expansile lobular or nodular lobulated growth and the alternate hypercellular and hypocellular areas of myoepithelial cells seen in MECAs. Compared with MECAs, cellular PAs showed more heterogenous arrangement of ductal, myoepithelial and stromal components (Figure 3). There was no nuclear pleomorphism/atypia, significant mitotic activity, or tumor necrosis identified in cellular PAs.

Figure 3. Kaplan Meier Curves for disease specific survival (A), recurrence free survival (B) and distant metastasis (DM) free survival (C).

Log rank test showed that MECA is associated with significantly decreased recurrence free survival and DM free survival compared with cellular PA, p=0.001 and 0.013, respectively.

MECA: myoepithelial carcinoma; PA: pleomorphic adenoma.

In contrast to the frequent local and distant recurrences observed in MECAs’ patients, among the 19 cellular PA patients with available follow up (median=23 months), none had distant metastasis and only one (6%) developed recurrent tumor in the same location, 54 months after the initial diagnosis, while the remaining 18 patients were disease-free at last follow up. The cellular PA that recurred had originally a positive margin at the time of first resection. All patients with cellular PA were alive without evidence of disease at their last follow up.

On univariate analysis using Log rank test, MECA group was associated with significant higher risk of recurrence (81% in MECAs and 6% in cellular PAs, p=0.001) and distant metastasis (31% in MECAs and 0% in cellular PAs, p=0.018) (Figure 3). The disease-specific survival did not differ significantly between patients with MECA and cellular PA (Log rank test, p=0.288). No regional lymph node metastasis was reported in neither of the cellular PA and MECA groups, including 28 patients with lymph node sampling at the time of initial surgeries (MECA: n=6 and cellular PA: n=22) and 35 patients with clinical follow up (MECA: n=16 and cellular PA: n=19).

Discussion

MECA is a histologically challenging entity with diverse histology and heterogenous morphologic and immunohistochemical features. This tumor can arise de novo or in association with a PA component (MECA ex-PA). Distinguishing MECA from its benign mimics is important since the former is associated with adverse outcome even when it is intracapsular or minimally invasive MECA ex-PA.

To date, only a handful of large case series have been published describing the histologic features and prognosis of this tumor (3, 4, 6–9, 14). The clinical outcomes of MECA reported in these studies are summarized in Table 3. Clinically, MECA seems to be relatively aggressive; it is associated with high risk of recurrence (cumulative recurrence rate of 43%, ranging from 0% to 59%), including local recurrence (cumulative rate of 35%, ranging from 0% to 59%), lymph node metastases (cumulative rate of 21%, ranging from 13-41%) and distant metastases (cumulative rate of 22%, ranging from 0 to 38%). In addition, patients with MECA appear to have a non-negligible risk of disease-associated mortality of 15% (range: 0-44%). In concordance with previous publications, MECA was associated with a high risk (31%) of distant metastasis and disease-related death (13%) in the current study. In contrast, its greater closest histologic mimic (cellular PA) did not show any distant metastasis and all patients were alive without evidence of disease. Moreover, in this study, 12 out 14 (85%) MECA patients who developed recurrences (8 local and 5 distant) had an initial benign diagnosis, including 7 PAs, 3 myoepitheliomas, and 2 atypical PAs. Although local recurrence by itself is not indicative of malignant behavior, the high frequency of local recurrence in MECAs (33%) compared with the local recurrence rate (6%) in cellular PAs suggests that MECA may be associated with more aggressive clinical behavior. Additionally, the same histologic criteria used to identify malignancy in MECAs with distant metastases can be applied to recognize malignancy in tumors without local and/or distant recurrences. Given the apparent diagnostic difficulty and the drastically different clinical outcomes between MECA and PA, it is crucial to recognize the pathologic features of MECA to distinguish this tumor from benign neoplasms in order to offer the proper management for these patients.

Table 3.

Literature review: clinical outcome of MECA.

Reference	N	Nodal metastasis	Recurrence (all)	Local recurrence	Distant metastasis	DOD
Nagao 1998 (14)	10	NA	2/9 (22%)	NA	3/9 (33%)	4/9 (44%)
Savera 2000 (4)	25	2/16 (13%)	10/17 (59%)	10/17 (59%)	6/16 (38%)	5/17 (29%)
Yang 2010 (6)	7	1/7 (14%)	0/5 (0%)	0/5 (0%)	0/5 (0%)	0/5 (0%)
Kane 2010 (7)	51	7/17 (41%)	21/51 (41%)	18/51 (35%)	3/51 (11%)	0/51 (0%)
Wang 2015 (8)	29	7/25 (28%)	11/27 (47%)	NA	7/25 (27%)	2/23 (9%)
Kong 2015 (3)	48	4/40 (8%)	NA	9/44 (20%)	12/44 (27%)	NA
Skalova 2015 (9)	25	5/21 (24%)	12/21 (57%)	11/21 (52%)	7/21 (33%)	8/21 (38%)
Total (from all previous studies)	195	26/126 (21%)	56/130 (43%)	48/138 (35%)	38/171 (22%)	19/126 (15%)
Current study	21	0/21 (0%)	14/16 (87%)	10/16 (63%)	5/16 (31%)	2/16 (13%)

The cohort in the present study was selected to reflect the difficulty that is encountered by pathologists in diagnosing MECA, as all studied MECAs were histologically challenging or initially misinterpreted as benign neoplasms. Such selection was intentionally done to emphasize the importance of differentiating MECA from its benign mimics and in order to characterize the histologic features that could be useful to identify malignancy in these tumors. Histologically, the two most helpful features in recognizing the malignant nature of MECA, distinguishing it from PA or myoepithelioma, were the nodular lobulated uniformly cellular growth of myoepithelial cells (identified in all studied cases) and the zonal cellular distribution (identified in 76% of studied cases) (Figures 1–2). These features were not demonstrated in any of the cellular PAs included in this study.

The zonal cellular distribution consists of a hypercellular peripheral rim surrounding a hypocellular center. The myoepithelial cells in the hypercellular rim can be epithelioid or spindled and the hypocellular center might vary in volume and can be necrotic, myxoid or hyalinized. Only rarely, MECA shows infiltrative destructive borders invading the adjacent tissue. In contrast, it often displays a pushing pattern at its periphery with a nodular growth with lobulated borders, especially in case of MECA ex-PA. We interpret this nodular lobulated growth as an expression of invasive pattern even when confined to the capsule of the tumor. Recognizing this expansile nodular lobulated pattern as a malignant invasive growth helps differentiate MECA from PA or myoepithelioma.

As mentioned above, one clue to the diagnosis of MECA is the uniform monotonous cytology of the myoepithelial cells in the cellular invasive nodules, perhaps reflecting their clonal growth and malignant transformation. Of note, benign protrusions or satellite nodules seen in PA outside the capsule should be distinguished from the malignant nodular pattern of MECA. These benign protrusions/nodules are typically seen in hypocellular myxoid PA. Furthermore, finding the three histologic elements of PA (ducts, myoepithelial cells and stroma), the heterogenous arrangement of these elements within the tumor, and the lack of the expansile myoepithelial cellular growth can help the pathologist in favoring a benign PA. Similarly, recurrent PA can exhibit multiple tumor nodules; however, it often has a miliary pattern of hypocellular nodules showing a typical PA morphology.

Interestingly, most of the misinterpreted cases in this study were found to be MECA ex-PA (13/18, 72%) which could be related to the difficulty in finding the malignant myoepithelial component. This can result in misclassifying the whole tumor as a benign PA. This seems to be a particular vexing diagnostic issue in the absence of high-grade features and when the PA component is intermixed with the malignant myoepithelial component, a feature that was observed in the majority of MECAs ex-PA in the current study.

In the third edition of WHO classification (15), cellular PA was recognized as a PA in which “the epithelial component forms the bulk of the tumor”. In the current study, we defined PA as cellular when the tumor showed a predominant (>50%) epithelial or myoepithelial component. A cohort of cellular PAs was chosen as a control group since cellular PA is the great histologic mimic of MECA; nevertheless cellularity has not been reported to carry a prognostic significance in PA (15).

Atypical PA (or PA with atypical features) is not a well-defined diagnosis and was not recognized as an entity in the WHO classification (1, 15). However, atypical features in PA have been reported in the literature to include: prominent zone of hyalinization, nuclear atypia/pleomorphism, elevated mitotic activity, atypical mitoses, hypercellularity, prominent central nucleoli, necrosis, vascular invasion, and capsular violation (16–22). Interestingly, prominent hyalinized stroma and elevated mitotic activity were reported to be prognostically relevant and we believe that these features could be histologic manifestations of malignancy (16).

Although there has been a controversy in earlier studies regarding the difference in clinical outcome with regard to the presence or absence of a PA component in MECA (3, 4, 14, 23) we have reported that MECA ex-PA correlated significantly with worse clinical behavior compared to de novo MECA (3) and that MECA type of CA ex-PA can recur and cause death even when it is intracapsular or minimally invasive (3). Similarly, in the current study, we found that the majority of recurrent MECAs (71%) were MECA ex-PA and mostly (80%) intracapsular or minimally invasive tumors. Moreover, the five patients who had distant metastases and the two patients who died from their diseases (one had initial misinterpretation of PA), had MECA ex-PA (Figure 1). These findings support the virulent nature of MECA ex-PA and highlight the significance of differentiating it from its benign mimics (3, 11).

By immunohistochemistry, expression for a cytokeratin and at least one of the myoepithelial markers, including S100, smooth muscle actin, GFAP, calponin, and p63, is typically required to establish myoepithelial differentiation. However, a variable staining pattern for myoepithelial markers can be found; thus, a panel of several myoepithelial markers should be always performed to confirm the myoepithelial origin when the diagnosis of MECA is histologically suspected. A diffuse S100 immuno-expression, although not totally specific, was found in all tested MECAs in this study and was reported before in the majority of MECAs of salivary glands; therefore, S100 should be always included in the immunohistochemical panel of MECA. Furthermore, immunohistochemistry can be utilized to highlight tumor architecture, identify the expansile lobulated nodular pattern within the tumor and show the distribution of ducts. Ki-67 immunostaining does not appear useful in recognizing malignancy in MECA, since all tested MECAs in the current study had a proliferative rate of ≤ 10%.

At the molecular level, PLAG1 and HMGA2 rearrangements have been reported to be the most common genetic events in both PA and CA ex-PA including MECA type (10, 24, 25). Hence, using FISH to detect PLAG1 and HMGA2 rearrangements might be valuable confirming the presence of the PA component but not differentiating benign from malignant. However, in a recent study by Dalin et al. describing the molecular alterations in MECA, TGFBR3-PLAG1 fusion was reported to be a potentially specific molecular marker for MECA, found in 15% of studied MECAs but not in PA or other benign salivary gland tumors (26). Thus, finding TGFBR3-PLAG1 fusion can be useful as a diagnostic maker of malignancy when the diagnosis of MECA is histologically suspected. In the same study, frequent oncogenic fusions involving PLAG1 were found in 53% of studied MECAs including cases that were histologically classified as de novo. Interestingly, the authors have also identified that high copy number alterations are more common in MECA ex-PA compared to de novo tumors, and this was found to correlate with poorer prognosis, supporting the aggressive nature of MECA ex-PA (26).

In conclusion, MECA is a challenging entity and may be easily overlooked and misclassified as a benign salivary gland neoplasm, especially when it is histologically bland. Cellular uniform myoepithelial growth with an expansile lobulated nodular pattern and zonal cellular distribution (i.e. peripheral hypercellular area surrounding a hypocellular center) are common characteristics of MECA and may serve as diagnostic histologic features to distinguish MECA from its benign mimics. Such distinction seems important as MECA can have adverse outcomes with a significant risk of distant and local recurrence even when it is intracapsular or minimally invasive MECA ex-PA.